reatom-feedback-loop

What problem does it solve?

This skill helps you correct and improve Reatom guidance when earlier answers caused the agent to deviate, produce wrong process, or miss key execution gates, so the same failure mode is less likely to repeat.

Core Features & Use Cases

• Process correction first: prioritizes the user’s immediate correction before discussing broader skill improvements.
• Reusable lesson distillation: converts the failure into concrete, skill-oriented edits (gates, examples, evaluation prompts) instead of blaming the agent’s memory.
• Failure-mode driven improvements: analyzes instruction-shape failures (what failed to constrain behavior) and places preventive rules in the right always-loaded or referenced guidance locations.

Quick Start

Ask an AI to run the Reatom feedback loop on your correction request, explicitly identifying which instruction or ordering constraint failed and proposing specific skill text, gate placement, or eval changes to prevent recurrence.